Worth separating two things that post #28 runs together.
Adding what did not work for me on non-inferiority margins, since the failures never get written up and they are half the useful information.
11 likes 7mo
Evidence · Trials · continued
Non-inferiority margins: how they are chosen and how they are abused posts 31–47
This is a continuation of a long topic, addressed by post number rather than by page. Start at post 1.
This follows post #31 rather than contradicting it.
Absolute and relative effects answer different questions. Write down the event rate in each arm and the difference between them; everything quotable is derived from those two numbers.
3 likes 7mo
Birkeland, post #13: Following, with nothing to contribute beyond having asked the same thing elsewhere. Go to post
The claim about non-inferiority margins upthread is stronger than its source supports. I have read the source. The source says "associated with" and the post says "causes".
My understanding of non-inferiority margins is a few years old and may have been superseded. If it has been, I would genuinely like to know rather than keep repeating it.
7 likes 6mo
c.lundgren, post #22: Absolute and relative effects answer different questions. Write down the event rate in each arm and the difference between them; everything quotable is derived from those two numbers. Stating my assumptions rather than smuggling them in. Go to post
Picking up post #35: that is the part I would want checked first.
Absolute numbers, not just relative: a 30% relative reduction tells you the ratio but not the practical magnitude. The event rate in each arm and the difference between them tells you how many people benefit.
I would put the burden of proof on the interesting explanation, not the dull one.
buffer_sheet, post #11: On post #9 — agreed on the reasoning, with one qualification. Non-inferiority margins has a well-known answer and a correct answer, and the interesting work is establishing that they are the same. Nobody has done that here yet. Go to post
I had written a reply contradicting post #35 and deleted it. Here is what survived.
The bit of non-inferiority margins that nobody enjoys is that the answer changes depending on what you are trying to decide with it. Say what the decision is and the thread will converge.
Risk of bias: structured appraisal of internal validity. Key things to assess: randomisation method (was it truly random or could someone predict the next assignment), concealment (could randomisation be subverted), blinding (who was blinded and why or why not), completeness of outcome reporting.
23 likes 6mo
A treatment-policy estimand asks what happens to people assigned to a strategy, including those who abandon it. A hypothetical estimand asks what would have happened had everyone continued. Both are legitimate and they give different numbers.
A partial answer, offered because a partial answer beats none.
11 likes 6mo
A definition problem is doing most of the work in this non-inferiority margins discussion. Once the term is pinned down I suspect the disagreement mostly goes away and what is left is small.
15 likes 6mo
d.oyelaran, post #4: Post #2 put the caveat in the right place and I want to underline it. Absolute numbers, not just relative: a 30% relative reduction tells you the ratio but not the practical magnitude. The event rate in each arm and the difference between them tells you how many people benefit. It is worth checking rather than assuming, which costs… Go to post
Post #38 describes the usual case. This is about the unusual one.
Non-inferiority margins are chosen, and the choice is an argument rather than a fact. A wide margin can make a worse treatment look acceptable.
Genuinely open to being wrong about this one.
Risk of bias: structured appraisal of internal validity. Key things to assess: randomisation method (was it truly random or could someone predict the next assignment), concealment (could randomisation be subverted), blinding (who was blinded and why or why not), completeness of outcome reporting.
The part I am sure of is shorter than the part I have written.
0 likes 6mo
Pre-specification is the property that makes a primary endpoint trustworthy. An endpoint chosen after seeing the data can be the best endpoint in the world and it no longer carries the same guarantee.
10 likes 6mo
Knowlton, post #9: On non-inferiority margins the community has more anecdote than the confidence in this thread implies, and I include my own contribution in that. Go to post
Building on post #45 rather than restating it.
An observation about non-inferiority margins that I cannot explain and am posting anyway, on the principle that unexplained observations are more useful public than private.
This topic was referenced in
- Interim analyses and stopping rules — does this still hold?Evidence › Trials · 34 replies
- Revisiting: Primary endpoint hierarchies and why order mattersEvidence › Trials · 25 replies
Promoted into the documentation commons. The content of this topic is maintained at STEP 8 — trial digest, with named maintainers and a review date. The promotion was discussed in doc review. Corrections are best raised against the document, which is the version that gets kept current.
Suggested topics
| Topic | Participants | Replies | Views | Activity |
|---|---|---|---|---|
|
[2026 update] Reading a supplementary appendix and finding the interesting part
On the subject in the title: Reading a supplementary appendix and finding the interesting part Working notes rather than a conclusion. Collecting what is known about Reading a supplementary appendix in one…
|
+39 | 44 | 8.2k | 5mo |
|
What a statistical analysis plan adds that the paper does not
What a statistical analysis plan adds that the paper does not — that is the question, and I have not found it answered plainly anywhere I have looked. Statistical analysis plan, and specifically the version…
|
2 | 6.6k | 10h | |
|
Adjudicated events and why the definition matters
On the subject in the title: Adjudicated events and why the definition matters Working notes rather than a conclusion. A follow-up question about adjudicated events that I did not know to ask the first time.…
|
2 | 4.5k | 2mo | |
|
Trial registration and comparing the protocol with the paper
Trial registration and comparing the protocol with the paper — setting out what I have, and where I think it stops being reliable. I have spent a fortnight trying to pin trial registration down and I want to…
|
+38 | 44 | 4.8k | 10mo |
|
[2026 update] Composite endpoints and the component doing the work
On the subject in the title: Composite endpoints and the component doing the work Working notes rather than a conclusion. Comparing SURMOUNT-1 ( N Engl J Med , 2022) with SURPASS-4 ( Lancet , 2021) and…
|
+76 | 87 | 994 | 3mo |
Related topics — sharing the tags open label, surrogate endpoints, estimand
| Topic | Participants | Replies | Views | Activity |
|---|---|---|---|---|
|
Composite endpoints and the component doing the work
Composite endpoints and the component doing the work Writing it up because I had to work it out twice and would rather nobody else did. Collecting what is known about composite endpoints in one place, because…
|
+3 | 7 | 33k | 13h |
|
Journal club: SUSTAIN 6 and its retinopathy signal
On the subject in the title: Journal club: SUSTAIN 6 and its retinopathy signal Working notes rather than a conclusion. Reading SUSTAIN 6 ( N Engl J Med , 2016) for the population rather than the effect,…
|
+110 | 122 | 46k | 1d |
|
Journal club: FLOW and the renal composite, component by component
On the subject in the title: Journal club: FLOW and the renal composite, component by component Working notes rather than a conclusion. Reading FLOW ( N Engl J Med , 2024) for the population rather than the…
|
+84 | 91 | 18k | 2y |
|
[2026 update] Journal club: PIONEER 6 as a safety trial
Journal club: PIONEER 6 as a safety trial Writing it up because I had to work it out twice and would rather nobody else did. Bringing a claim rather than a paper, because the claim is the thing circulating.…
|
+14 | 18 | 55k | 5mo |
|
Interim analyses and stopping rules
Interim analyses and stopping rules Writing it up because I had to work it out twice and would rather nobody else did. Asking about interim analyses and stopping rules on behalf of the question I keep seeing…
|
+25 | 29 | 35k | 16d |